Updated Sep 13
Nov 8, 2026
AI Auditors Catch Agent Misbehavior But Flood Reviewers With False Alarms
A new study finds frontier AI models can catch nearly all reward hacking in agent logs, but false alarms pile up so fast at scale that human reviewers would drown in noise — getting 3,000 false warnings for every 2,000 real incidents.
ai auditingai agentsreward hacking
Nov 8, 2026
Elvis Estate and Eminem Publisher Sue Suno in Escalating AI Music Copyright War
The rights holders for Elvis Presley and Eminem's publisher Eight Mile Style filed separate federal lawsuits against AI music generator Suno on Friday, alleging copyright infringement and violations of Tennessee's ELVIS Act. The twin filings mark one of the first major tests of the 2024 voice-cloning law and escalate an already intense legal battle between the music industry and AI companies.
sunoai-musiccopyright
Nov 8, 2026
Revolut Swaps Claude Code for Chinese GLM, Cuts Coding Costs 90%
Revolut has moved roughly 2,000 engineers from Anthropic's Claude Code and OpenAI's Codex to a self-hosted instance of GLM 5.3, an open-weight model from Beijing-based Z.ai, cutting AI-assisted coding costs by approximately 90%. David Abitbol, who leads AI operations and infrastructure at Revolut, shared the migration story at the AI Engineer conference in New York, detailing how the fintech tuned the model from a "terrible" out-of-box experience of 14-second latency down to 1.6 seconds. The move underscores a growing trend: well-funded engineering organizations are betting on self-hosted open-weight models over per-seat SaaS subscriptions for AI coding tools.
revolutglmai-coding-costs
Related News
Nov 2, 2026
Google’s Africa learning hub separates offline lessons from AI assistance
The November 2 launch offers downloadable modules and SCOUT guidance, without stating that its assistant works offline.
AI educationGoogle
Oct 29, 2026
Google’s GenAI Python SDK fixes request headers becoming client defaults
Version 2.34.0 changes how request headers are copied and drops Python 3.10 support. The linked patch includes regression cases for timeouts and credentials reused across SDK operations.
GoogleGenAI SDKPython
Oct 29, 2026
Anthropic’s expanded science credits follow different research-access routes
Anthropic’s October 28 expansion distributes Claude credits through funders, inside research institutions and directly to specific initiatives. Researchers need to identify which route can support their project.
AnthropicClaudeAI for Science